应用独立成分分析和小波分解对木材声发射信号的析取  

Extraction of Wood Acoustic Emission Signal by Independent Component analysis and Wavelet Decomposition

作　　者：罗蕊寒 方赛银[2] 丁锐 赖菲 王明华 李晓崧 罗廷芳[2] 李明 Luo Ruihan;Fang Saiyin;Ding Rui;Lai Fei;Wang Minghua;Li Xiaosong;Luo Yanfang;Li Ming(Southwest Forestry University,Yunnan 650224,P.R.China;Anhui Polytechnic University)

机构地区：[1]西南林业大学机械与交通学院,昆明650224 [2]西南林业大学,昆明650224 [3]安徽工程大学电气工程学院、高端装备先进感知与智能控制教育部重点实验室(安徽工程大学)

出　　处：《东北林业大学学报》2022年第2期83-87,共5页Journal of Northeast Forestry University

基　　金：国家自然科学基金项目(31760182、31100424);云南省教育厅科学研究基金项目(2020Y0376);云南省教育厅资助性项目(2017ZZX215)。

摘　　要：选取常温气干状态表面无缺陷的樟子松(Pinus sylvestris var.mongolica Litv.)实木为试验材料,制成长800 mm、宽60 mm、厚30 mm的试件。使用UTM5105型万能力学试验机对试件进行破坏性试验,以500 kHz的采样频率采集木材三点弯曲试验产生的声发射(AE)信号,截取试验后期幅值无显著变化的一段原始信号作为研究对象。采用依据负熵最大化的快速独立成分分析(FastICA)盲源分离算法将原始信号分离成噪声和声发射信号,再对分离后的声发射信号进行5层小波分解后重构声发射信号波形;对重构声发射信号进行频域分析,通过与已知声发射信号的频域特征比较,验证信号析取的有效性。结果表明:构建的依据独立成分分析和小波分解(FastICA-Wavelet)的声发射信号析取方法,能够从混有声发射信号的类噪声信号中分解出声发射信号,利用小波分解能够进一步降低非独立噪声成分的影响。The solid wood of Pinus sylvestris var.mongolica Litv.with no surface defects in the air-dried state at room temperature was selected as the test material,and the specimens with a length of 800 mm,a width of 60 mm and a thickness of 30 mm were made.The UTM5105 universal chemical testing machine was used to conduct destructive tests on the specimens.The acoustic emission(AE)signals generated by the three-point bending test of wood were collected at a sampling frequency of 500 kHz,and a segment of the original signal with no significant change in amplitude in the later stage of the test was intercepted as the research object.The original signal was separated into noise and acoustic emission signal by using Fast Independent Component Analysis(FastICA)blind source separation algorithm based on negentropy maximization,and then the separated acoustic emission signal was decomposed with five layers of wavelet to reconstruct the acoustic emission signal waveform.The reconstructed acoustic emission signal was analyzed in the frequency domain,and the validity of the signal extraction is verified by comparing it with the frequency domain characteristics of the known acoustic emission signal.The results show that the acoustic emission signal extraction method based on independent component analysis and wavelet decomposition(FastICA-Wavelet)can decompose the acoustic emission signal from the noise like signal mixed with acoustic emission signal,and the influence of non-independent noise components can be further reduced by wavelet decomposition.

关 键 词：樟子松 木材 声发射 信号析取 

分 类 号：S781.3[农业科学—木材科学与技术] TN911.7[农业科学—林学]